Learnijoy
ResearchAI ResearchAI News & Tools

XAI in Computational Pathology: A Formal Review and Roadmap

Shubham Innani, Suhang You, Adam Shephard, Bhakti Baheti, Francesco Ciompi, Joe Yeong, Nasir Rajpoot, Michael Feldman, Solene Florence Kammerer-Jacquet, Dimitrios Makris, Geert Litjens, Anne L. Martel, Jana Lipkova, April Khademi, Spyridon Bakas, for the MICCAI SIG-CompPath· September 1, 2026 View original

Key takeaways

  • XAI is crucial for building trust and enabling verification of AI in computational pathology.
  • A new pathology-centric vocabulary and taxonomy formalize XAI methods in CompPath.
  • A task-driven framework maps clinical questions to recommended XAI methods and evaluation.
  • Addressing identified gaps is essential for advancing XAI towards clinical deployment and regulatory acceptance.

Who benefits

HealthcarePharmaceuticalsMedical DevicesAI DevelopmentRegulatory Affairs

Summary

This review formalizes Explainable AI (XAI) methods in computational pathology (CompPath) by introducing a pathology-centric vocabulary, developing a taxonomy of methods, and establishing a task-driven framework. It identifies key gaps between current XAI capabilities and clinical deployment, proposing actionable steps for advancement.

The field of computational pathology (CompPath) is being revolutionized by AI algorithms that assist in diagnosis, prognosis, and treatment prediction from gigapixel whole-slide images. However, widespread clinical adoption is hindered by concerns regarding safety, accountability, and regulatory oversight, especially in high-stakes medical environments. Explainable AI (XAI) systems offer a promising solution to build trust and enable verification, but the existing literature is fragmented by inconsistent terminology and ad hoc validation. This comprehensive review aims to formalize XAI methods within CompPath. It introduces a pathology-centric vocabulary with seven core terms, providing a standardized language for the field. Furthermore, it develops a robust taxonomy that categorizes methodological families across three orthogonal axes: stage, type, and scope of explanation. A task-driven framework is also established, mapping five critical clinical questions to recommended XAI methods, their evaluation strategies, and appropriate deployment contexts. The review identifies five key gaps that currently exist between XAI capabilities and their full clinical deployment. To bridge these gaps, it proposes actionable steps designed to advance XAI for CompPath, ultimately fostering greater trust, regulatory compliance, and clinical utility for AI in pathology.

Why it matters

For professionals in healthcare, AI development, and regulatory affairs, this review provides a critical framework for understanding, evaluating, and deploying Explainable AI in computational pathology, accelerating safe and effective clinical adoption.

How to implement this in your domain

  1. 1Adopt the proposed pathology-centric XAI vocabulary to standardize communication and understanding within multidisciplinary teams.
  2. 2Utilize the XAI taxonomy to systematically evaluate and select appropriate explanation methods for specific clinical AI applications.
  3. 3Integrate the task-driven framework into AI development pipelines to ensure XAI methods directly address clinical questions and needs.
  4. 4Collaborate with regulatory bodies and clinicians to address identified gaps and develop robust validation strategies for XAI in high-stakes medical AI.

Original post by Shubham Innani, Suhang You, Adam Shephard, Bhakti Baheti, Francesco Ciompi, Joe Yeong, Nasir Rajpoot, Michael Feldman, Solene Florence Kammerer-Jacquet, Dimitrios Makris, Geert Litjens, Anne L. Martel, Jana Lipkova, April Khademi, Spyridon Bakas, for the MICCAI SIG-CompPath

"arXiv:2608.28820v1 Announce Type: new Abstract: Computational pathology (CompPath) is transforming medicine by leveraging artificial intelligence (AI) algorithms to support diagnosis, prognosis, and treatment prediction from gigapixel whole-slide images. Clinical adoption is prog…"

View on X

Originally posted by Shubham Innani, Suhang You, Adam Shephard, Bhakti Baheti, Francesco Ciompi, Joe Yeong, Nasir Rajpoot, Michael Feldman, Solene Florence Kammerer-Jacquet, Dimitrios Makris, Geert Litjens, Anne L. Martel, Jana Lipkova, April Khademi, Spyridon Bakas, for the MICCAI SIG-CompPath on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses

More in AI Research

AI Engineering & DevToolsAI Research

PAC-LLM Forecasts Chaotic Time Series with LLMs

PAC-LLM is a phase-space-aware adaptive fusion framework that leverages Large Language Models (LLMs) to forecast long-term chaotic time series, even with limited short-term observations. It integrates learned phase-space features and textual information to enhance LLM forecasting capacity.

Yuhang Yao, Bohan JiangSep 1, 2026
AI Engineering & DevToolsAI Research

Event-Triggered Control for Networked Systems with Delays

This paper proposes an efficient control framework with an asynchronous event-triggered mechanism for networked systems, accounting for computational delays in online learning. It guarantees control performance while optimizing communication and computation resources.

Xiaobing Dai, Armin Lederer, Zewen Yang, Sihua Zhang, Lu Wan, Yang Tang, Sandra HircheSep 1, 2026
AI Engineering & DevToolsAI Research

HoopMind: AI System for Real-Time Basketball Strategy

HoopMind is a real-time neural game-tree system that fuses public basketball data to model half-court possessions as sequential games, providing opponent-aware possession planning. It offers a scouting planner and playable simulator for strategic analysis.

Yibo Gong, Cong Guo, Jiacheng DingSep 1, 2026